Preprocessing To Be Considered For MR and CT Images Containing Tumors

Abstract

Image pre-processing is an important and challenging factor in the computer-aided diagnostic systems. In medical image processing and especially in tumor segmentation task it is very important to pre- process the image so that segmentation and feature extraction algorithms work correctly. Proper detection and segmentation of the tumor leads to exact extraction of features and classification of those tumors. The accurate tumor segmentation is possible if image is pre-processed as per image size and quality. This paper describes the pre-processing method consisting of two phases. In the first phase we remove the film artifact by using median filter. In the second phase we introduce an algorithm that uses morphological operations to remove unwanted skull/ribcage portion. This reduces the false positive results in the later stages of processing in the computer- aided diagnostic systems .Both algorithms are applied on MR/CT images of Brain having tumors and, CT images of Thorax and Abdomen having tumors. The second algorithm has an effect of skull portion removal in Brain images and, effect of ribcage portion removal in CT images of Thorax and Abdomen. The results are superior to those of tracking algorithms considering the retention of regions of interests and the film artifacts removal. This preprocessing will also reduce the over segmentation problem in further processing still retaining the tumors.